Speed lab

See this page's own speed, measured live in your browser, and what slows a site down.

WhatsApp me

I reply the same working day, in UK hours.

Or call 07935 996850 or book a call

5.0 from 31 Google reviews (checked Nov 2026)

Lab 07 of 09

Core Web Vitals
Performance
Real measurements

Try it

Lab 07 of 09

Your site, tested by Google

Test your own site

Type your web address and Google's PageSpeed Insights will load the page and measure it, the way it would on a phone. It can take up to a minute.

Part 1. Real measurements

This page, measured in your browser

These are your browser's own figures for this page, on your device and your connection, taken while you're here. Nothing is sent anywhere. Someone else will see different numbers.

LCPLargest Contentful Paint

Core Web Vital

Measuring

When the biggest thing on screen finished drawing.

Good: 2.50 s or less. Poor: over 4.00 s.

CLSCumulative Layout Shift

Core Web Vital

Measuring

How much the page jumped about while you were reading it.

Good: 0.10 or less. Poor: over 0.25.

INPInteraction to Next Paint

Core Web Vital

Waiting for a tap

How long the page took to react when you tapped, clicked or typed.

Tap, click or type anywhere on the page to measure this. Or try these:

The slow one keeps your browser busy for 300 ms on purpose, like a heavy script would.

Good: 200 ms or less. Poor: over 500 ms.

FCPFirst Contentful Paint

Measuring

When the first words or picture appeared.

Good: 1.80 s or less. Poor: over 3.00 s.

TTFBTime to First Byte

Measuring

How long the server took to start sending the page.

Good: 800 ms or less. Poor: over 1.80 s.

Where the time went

Measuring...

Page weight

Measuring...

Good and poor lines from web.dev, checked 5 November 2026: LCP, LCP poor line, CLS, INP, FCP, TTFB. LCP, CLS and INP are the Core Web Vitals; FCP and TTFB help explain them. web.dev suggests judging a site on the 75th percentile of real visits. This is just one visit: yours.

Simulation Part 2. What slows a site down

Flip the usual culprits and watch it load

A made-up example page, a joiner in Sunderland, on a made-up phone connection. Switch each problem on or off and the page reloads in front of you. The numbers come from a simple model of how these problems behave. They're illustrative, not measurements.

Straight off the camera at 3.4 MB, and nothing tells the browser it's important.
Loaded from another server, and the words stay hidden until it arrives.
When it lands, everything under it jumps down the screen.
900 KB from a third party, then long tasks that keep the browser too busy to respond.

Simulated Results for this example

LCP4.56 sPoor

FCP822 msGood

CLS0.27Poor

INP510 msPoor

Page weight 4.91 MB

Filmstrip

One frame every 500 ms of the simulated load. Tap a frame to jump to it.

Timeline

What the example browser downloads, and when. Striped bars hold up the first paint; the bottom row is the browser's own work.

Page HTML 30 KB

Styles 25 KB

Web font (blocking) 390 KB

Photo (unoptimised) 3,400 KB

Heavy script 900 KB

Advert 160 KB

Browser busy long tasks

Part 3. Measured, not simulated

My own site, before and after

The same test on my old WordPress site and on this new build. The bar is LCP, and shorter is better. The two thin lines on each bar are web.dev's good (2.5 s) and poor (4 s) lines.

Load times of my old site and the new build, on a throttled phone, 23 October 2026
Page	Old site (WordPress, live)	New site (built 22 Oct)
Home page	LCP 21.2 sFCP 3.0 s	LCP 1.75 sFCP 1.75 s
My blog post on website costs	LCP 12.3 sFCP 3.4 s	LCP 0.9 sFCP 0.9 s
A second page (cached)	LCP 5.5 sFCP 1.1 s	LCP 0.6 sFCP 0.6 s
Website design page	LCP 3.5 sFCP 2.6 s	Not measured
Care+ page	LCP 6.0 sFCP 6.0 s	Not measured

Measured on 23 October 2026 with Lighthouse's mobile settings (a throttled phone: 150 ms latency, 1.6 Mbps, 4× slower CPU), the median of 3 runs. The same method on both sites, so the numbers compare. The old site was the live WordPress site; the new one was the build from 22 October. Your own visit, in Part 1, will differ.